> ## Documentation Index
> Fetch the complete documentation index at: https://docs.colloqui.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Prompt examples

> Ready-to-use prompt patterns for common voice agent scenarios — pronouncing contact details, handling silence, and more.

These are practical prompt snippets you can drop into your agent's `prompt_config.prompt` to handle situations that come up on almost every voice agent. Each one solves a specific problem with tested wording.

## Pronouncing phone numbers

Models tend to read phone numbers as large integers ("four billion, one hundred fifty-eight million...") or rush through digits without pauses. Add this to your prompt to force digit-by-digit pronunciation:

```markdown theme={null}
When speaking a phone number, break it into individual digits
grouped with pauses:

- 4158923245 becomes "four one five — eight nine two — three two four five"
- Always insert a pause (dash) between groups of three or four digits
- If a caller asks you to repeat the number, use the same format every time
```

This works across all TTS providers. The dash characters create natural pauses in speech.

## Spelling out email addresses

Email addresses are tricky because they mix letters, symbols, and domain names. Instruct your agent to spell them out clearly:

```markdown theme={null}
When speaking an email address:

- Spell out the part before @ letter by letter: "j-o-h-n"
- Pronounce @ as "at"
- Spell out the domain if it's not a common word: "a-c-m-e"
- Say "dot" before the extension: "dot com", "dot net"

Example: john@acme.com becomes "j-o-h-n at a-c-m-e dot com"

If the caller asks you to repeat it, use the exact same format.
```

## Pronouncing website URLs

```markdown theme={null}
When speaking a website URL:

- Never say "HTTP" or "HTTPS" or "www" unless the caller needs
  the full URL for a technical reason
- If part of the domain is a word, say the word normally
- If part of the domain is letters, spell each one out
- Always say "dot" before the extension

Examples:
- nklaundry.com → "en-kay-laundry dot com"
- abctest.net → "ay-bee-see-test dot net"
```

## Speaking times and dates

Models default to written formats like "3:30 PM" which TTS can mangle. Be explicit:

```markdown theme={null}
When speaking times:

- 1:00 PM → "one PM"
- 3:30 PM → "three thirty PM"
- 8:45 AM → "eight forty-five AM"
- Always include AM or PM
- Never say "o'clock"

When speaking dates:

- 01/15 → "January fifteenth"
- 03/03/2025 → "March third, twenty twenty-five"
- Never use numeric date formats when speaking
```

## Handling silence and holds

When a caller says "hold on" or goes quiet, you usually don't want the agent to keep talking. How to handle this depends on your LLM provider.

### Standard approach

Add this to your prompt:

```markdown theme={null}
If the caller says "hold on", "one moment", "one second",
or similar:

- Say "Sure, take your time" and then wait silently
- Do not ask follow-up questions or fill the silence
- When the caller speaks again, respond normally
```

Pair this with your `prompt_config` settings — set `ai_speak_after_silence` to `true` and `ai_speak_wait_time` to a comfortable duration (3–5 seconds) so the agent gently re-engages if the silence goes on too long.

### Adjusting silence behaviour

The `tts_config.responsiveness` setting (0–1) controls how quickly the agent jumps in after the caller stops speaking. Lower values give callers more space; higher values make the agent respond faster. For calls where callers might need time to think (scheduling, providing account numbers), set this around 0.3–0.5.

## Handling interruptions

```markdown theme={null}
If the caller interrupts you mid-sentence:

- Stop speaking immediately and listen
- Do not repeat what you were saying unless the caller asks
- Respond to whatever the caller said
```

The `tts_config.interruption_sensitivity` setting (0–1) controls how easily callers can cut in. Higher values mean the agent yields more readily. For conversational agents, 0.7–0.8 works well.

## Confirming important information

When your agent collects critical details — booking times, phone numbers, spelling of names — always confirm:

```markdown theme={null}
After the caller provides any of the following, repeat it back
and ask for confirmation before proceeding:

- Phone numbers
- Email addresses
- Appointment dates or times
- Names (especially spelling)
- Order numbers

Example: "Just to confirm, that's j-o-h-n at a-c-m-e dot com — is that correct?"
```

## Handling "I don't know" gracefully

```markdown theme={null}
If you don't have enough information to answer a question:

- Say so honestly: "I don't have that information available right now"
- Do NOT make up an answer or guess
- Offer an alternative: transfer to a team member, take a message,
  or suggest the caller check a specific resource
```

## Keeping responses concise

The single most common problem with voice agents is responses that are too long. Add this as a hard constraint:

```markdown theme={null}
IMPORTANT: Keep every response to 1–2 sentences maximum.
The caller is listening on a phone, not reading a screen.
If you need to convey multiple pieces of information,
break them across separate turns and check for understanding
between each one.
```